Guinness vs Murphy's: what the LLMs actually say
Murphy's is having a year. UK value sales up 666%, total-trade volume up over 1,000%, its on-trade share doubled to 4.2% - the fastest-growing stout in Britain, riding the wake of the Guinness rationing crisis. But large language models were trained before most of that happened. So we asked a simple question: when ChatGPT, Perplexity and Gemini talk about Irish stout, have they noticed Murphy's yet - or is it still the Guinness show? We ran both brands through a Baseline mention scan to find out — the same method behind the Barry's vs Lyons tea scan and the Revolut vs Monzo banking scan.
Guinness still owns the conversation
Across the LLM answers and the sources they cite, Guinness pulled 306 mentions to Murphy's 141 - roughly 2.2× the visibility. As of June 2026, the real-world momentum has not reached the models: Guinness is still the default answer to "what's a good stout," and Murphy's surge is, for now, invisible to them.
But the underdog is better-liked
Here's the twist. Murphy's is mentioned far less - but when it is, the tone is warmer and cleaner. It edges Guinness on average sentiment and carries barely a third of the negativity. Guinness's ubiquity cuts both ways: more coverage, but also more criticism to absorb.
| Metric (as of 19 Jun 2026) | Guinness | Murphy's |
|---|---|---|
| Total mentions | 306 | 141 |
| Average sentiment | +0.25 | +0.28 |
| Positive | 50% | 47% |
| Neutral | 41% | 49% |
| Negative | 8.5% | 4.3% |
The scan's analyst pass put numbers to why. Guinness's negatives are concrete and reputational - the Guinness 0.0 contamination recall scored −0.8 on the brand's own Twitter, and the 1990 fraud scandal still surfaces in high-authority coverage. Murphy's, with only six negative mentions in the whole scan, simply has not accumulated a critical narrative yet. There's less to say about it - and less of it is bad.
Watch it move - updates weekly
The numbers above are a snapshot from the date on this post. The graph below is different: it re-draws itself straight from our platform every time a fresh weekly scan lands, so you can watch whether the models ever catch up to Murphy's real-world run. One datapoint today; a trend over time.
A headline vs a recommendation
The volume gap is only half of it. The two brands live in different corners of the web. Guinness surfaces as a news entity: the BBC, CNN and The Guardian account for a big slice of its mentions, much of it the rationing-crisis and recall coverage. Murphy's is word-of-mouth: Reddit, Facebook and retail sites like TotalWine. Guinness is something the press writes about; Murphy's is something drinkers and shops recommend.
The gap between the shelf and the model
There's a lag between what's happening in pubs and what the AI answers reflect. Murphy's is one of the fastest-moving brands in its category in the real world, yet to ChatGPT and friends it's still a footnote to Guinness - mentioned half as often, framed as the alternative rather than the story. For a challenger brand that's the whole opportunity: the models update as the web does, and the brands that show up in the sources LLMs cite are the ones that get recommended — a shift most brands aren't even tracking. The shelf has moved. The model hasn't - yet. The tracker above is how we'll know when it does.
Grounded in Baseline Labs platform data - Guinness scan #363 (template #1185), Murphy's scan #364 (template #1186), save #729. Live trackers refresh weekly via scheduled jobs #307 & #308 (auto-pause after 26 runs). Real-world sales figures from public 2026 trade-press reporting (The Grocer, The Drinks Business, Bar Magazine, Morning Advertiser).
Image (brand-treated from licensed original): Pint of stout by 55belgianbeerfan, CC0, via Wikimedia Commons.
